Epochal work by Nina G. Jablonski and George Chaplin of Department of Anthropology, California Academy of Sciences.
A standard two-sample t-test was
used to determine the significance of the
differences between the sexes at the same
locality. Separate analyses for the northern and southern hemispheres were not undertaken
because the small sizes of the datasets
that met the above criteria would
have rendered hemisphere-specific analyses
meaningless. Further, our objective here was
not to test the hypothesis that there is a
difference between the hemispheres in the
degree of differentiation between the two
sexes in skin reflectance.
The relationship of annual average UVMED
to skin reflectance
A correlation matrix was used to test the
strength of the relationships between annual
average UVMED, latitude and skin reflectance.
This analysis was undertaken
specifically to determine the strength of the
correlation of skin reflectance to UVMED
relative to its correlation with latitude. The
skin reflectance data used in this analysis
comprised those from indigenous populations,
all-sex samples combined, averaged
by country. For this analysis, latitude was
transformed to absolute latitude, i.e., the
absolute value of the latitude, which
expresses the angle of a location from the
Equator rather than relative north and
south. It is unclear from previous studies of
human skin pigmentation as to whether
workers utilized conventional latitude or
absolute latitude for purposes of correlation
or regression analyses. Transformation of
latitude to absolute latitude is necessary
because of the nonlinear relationship
between solar insolation and latitude. The
transformation to absolute latitude renders
this relationship more linear, and thus yields
much higher correlations.
The relationship between UVMED and
skin color reflectance was explored in
greater detail using a least squares regression.
For this analysis, the skin reflectance
data used were the same as those employed
in the correlation matrix analysis described
above. Separate regressions were developed
for data from each hemisphere, for each widely used skin reflectance filter (425, 545
and 685 nm).
Predicted vs. observed skin reflectances
In order to compare predicted vs. observed
values for the skin reflectances of indigenous
peoples, the largest available dataset for
observed skin reflectance (at 685 nm) was
used. It is important to note again that some
populations were represented by males only,
as both sexes combined, or as unspecified
sex, as discussed. Although these variations
in sexual classification of the raw data added
to the expected variance of the pooled dataset,
this problem could not be avoided.
Significant outliers from the regression line
were then identified.
In order to construct a map of predicted
skin colors for modern terrestrial environments,
a regression was computed between
annual average UVMED and the observed
skin reflectance. The observed reflectances
for indigenous populations were based on all
available data for a particular area or group.
A regression equation derived from a larger
data set including values for males, females,
both sexes and samples of unknown sex was
used:
Predicted skin color
=annual average UVMED
(X - 0·1088)+72·7483.
THE EVOLUTION OF HUMAN SKIN COLORATION, Part 10 - Study continues HERE.